MIXED GAME PRO

Experimental solver

How the solver works

CFR+, Nash equilibrium, and the numbers you see

  1. Build the situation.Your action path sets the pot, effective stacks, filtered ranges, draw choices, and player to act.
  2. Group similar hands.To keep the game tree practical, hands with similar strength are placed into buckets such as B24. Each row represents a class of hands, not one exact hand.
  3. Run CFR+ iterations.Counterfactual Regret Minimization repeatedly tests every available action, records the value lost by not choosing a better action, and updates both players' strategies. CFR+ removes negative regret to converge faster.
  4. Average the strategy.The percentages are action frequencies in the same situation. For example, 70% raise and 30% call is a deliberate mixed strategy—not a confidence score.
What does Nash mean?

A Nash equilibrium is a pair of strategies where neither player can gain EV by changing strategy alone. This solver approaches that point; the regret gap is its convergence estimate, and lower is better.

This is an abstracted experimental solve, not an exact full-card game solution. Deep, rare action lines and hands grouped inside one bucket can be less precise.

2-7 Single Draw (No Limit) heads-up solver.

Set the effective stack, choose the pre-draw betting line and draw declarations in playing order, then explore the derived post-draw strategy.

Review game rules Open equity calculator

Experimental solver · community supported

We do not have much budget to rent a faster server and develop more features.If this solver helps your game, please consider donating to faster compute + active development.
Donation address (ETH, USDC, USDT...)0x2f3fb5b9aef2bb2017c70981bcc69ba3267895c3

2-7 Single Draw (No Limit)

Solving the game tree…

The engine derives continuing ranges from the pre-draw size, applies both draw declarations, then runs CFR across the post-draw betting tree.

Pre-drawDrawPost-drawStrategy

Model boundary

A real equilibrium solve, with a deliberately narrow first tree.

01

Action-ranked ranges

Pat and draw ranges are ranked independently, so made 9, T, and J lows can remain in a realistic pat range while draw hands are ranked by their post-draw potential.

02

CFR+ strategy

The engine iterates every abstract hand matchup and returns average equilibrium action frequencies.

03

Convergence estimate

The reported regret gap tracks remaining positive regret across the solved strategy profile.